Abstract:
High-impedance fault (HIF) detection is crucial for maintaining the reliability and resiliency of microgrid systems. This
research presents an adaptive machine learning approach to enhance HIF detection and improve resiliency against the outage of
optimally placed phasor measurement units (PMUs) in microgrids. PMUs are strategically positioned in limited numbers across the
microgrid, considering their cost-effectiveness. When one of these PMUs encounters an outage, HIF detection becomes more complex
due to the critical information loss from the affected area. The proposed approach utilizes a combined framework of correlation
modelling, feature extraction using Hilbert-Huang Transformation (HHT), and Analysis of Variance (ANOVA). By leveraging
machine learning algorithms, the approach selects the most relevant features derived from Hilbert spectral analysis (HSA) to perform
tasks such as PMU outage detection, HIF detection, and classification during optimally placed PMU outage scenarios. The effectiveness
of the approach in enhancing resiliency for high-impedance fault (HIF) detection during PMU outage scenarios is demonstrated through
simulation studies conducted in MATLAB Simulink on microgrid systems.
